Added
- ResNet 18, 34, 50, 101, 152 classifiers with pretrained weights (pytorch)
- Xception classifier with pretrained weights (Sorbonne Labs)
- ResNet 18, 34, 50, 101, 152 backbones with pretrained weights (pytorch)
- Xception backbones with 8 and 16 middlw flow layers and pretrained weights (adapted from the classifier)
- U-Net model with 2D deconvolution (original) and bilinear upsampling (enhanced version)
- DeepLabV3 using RN50/101 or Xception backbones
- DeepLabV3+ using RN50/101 or Xception backbones
- First PiPy release